G. Translational Research Core 1: Health Disparities (Core Director. Andy J. Karter PhD) The following sections outline the three translational research cores of the HMORN-UCSF CDTR. All three will be administered, and their expertise disseminated, through the Administrative Core's Enrichment Program as describe above (section D). G.l. Importance of Core to Translational Research in Diabetes and NIDDK It may be incorrect to assume that the introduction of a novel health intervention into a new practice setting (Type 2 translation) will yield uniform benefits across all patient subgroups ("uniform benefit assumption"). An intervention proven to be efficacious in one setting might not have been tested in populations with vulnerable or disadvantaged subgroups, such as patients who are poor or less educated, have low health literacy, or belong to minority subgroups. Ensuring that novel interventions benefits all exposed, including vulnerable subgroups, requires monitoring and may require modified approaches or specially targeted efforts. The choice of which target population to apply the translational intervention (e.g., the whole population at risk, only those with elevated risk factors, or the vulnerable populations) may reduce or exacerbate health disparities.(107)'The investigators in this research core will stress the importance of monitoring group-specific effects from the start pt teach new translational intervention and encourage, when possible, conducting pilot studies powered to evaluate subgroup differences before fully implementing the intervention. Early detection of subgroup differences provides an important opportunity to tailor interventions (or make midstream modifications) to better suit those not initially benefiting. Sub-group differences in a translational intervention's effect can result from differences in the adoption of new interventions between subgroups, differing impacts of a new health care policy, and even clinical/biological variation in the effects of an intervention. Social factors such as health literacy, language proficiency, education, income and race/ethnicity may impact the effect of translating a new intervention into practice. The "Inverse Care Hypothesis"(108,109) proposes that the introduction of a new intervention can yield "further advantages for the already advantaged." Vulnerable populations may lack needed social and physical resources needed to avoid disease risk or adopt health-promoting behaviors. Thus, interventions may not be equally effective in all subgroups if the well-off (e.g. better educated, those with higher incomes or better "connected") take advantage of a new intervention more quickly or more widely, or if vulnerable populations experience barriers to access (e.g., financial or language-based) or are slower to adopt the new intervention, or if the intervention itself does not meet the needs of some subgroups. Thus, new interventions, even if conferring benefits to the entire population exposed (i.e., improvements for the "average" person), can unintentionally increase relative health disparities, leaving vulnerable subgroups further behind, with benefits accruing disproportionately to the well-off or risks accruing disproportionately to the vulnerable. The Diabetes Study of Northern California (DISTANCE) reported that patients with low health literacy (e.g. difficulty understanding health education materials or instructions) were less likely to adopt the use of a widely offered internet-based patient portal even if they had access to a computer.(110) We also have preliminary evidence that offering a financial incentive to use mail order prescription refills, while improving use of mail order refills overall and promoting adherence, had a lesser impact in vulnerable populations (minorities, lower income or less educated patients). A corollary question is whether some innovations might actually increase health risks for vulnerable populations. In DISTANCE, we found that diabetes patients who had poorly controlled blood glucose and were prescribed insulin were at greater risk of serious hypoglycemic events if they had low health literacy. (111) A population based study (NHANES) reported a growing disparity in diabetes-related mortality across education levels in the late 1980's until 2005, overall and within gender and race/ethnic subgroups.(112) Thus, there is evidence that innovations in health care, while improving overall health, may unintentionally lead to increased disparities. In contrast, there is a small but growing body of translational research which suggests that developing and/or tailoring innovative healthcare interventions with the explicit goal of reducing health disparities can level the playing field, and even "reverse the inverse care law".(109) The use of health communication technology to support self-management in diabetes and heart disease has been investigated by the group at the UCSF Center for Vulnerable Populations;they demonstrated that involving the target population in development, tailoring the intervention to meet the needs of vulnerable populations and the health providers and systems that disproportionately care for them, attending to the contextual demands of people's lives when implementing the intervention, can improve the acceptance of the intervention, improve care overall and yield disproportionate improvements among those with limited literacy or limited English proficiency. (113-115) The creation or implementation of health care policy may also provide differing benefits across social subgroups. In recent years, 38 states have passed legislation mandating that health plans provide coverage for diabetic supplies, with the intent to eliminate financial barriers to self-monitoring of blood glucose (SMBG). Three of the centers included in this CDTR application (Kaiser, UCSF and Harvard) reside in states that experienced these legislative mandates during 1990-2000. Kaiser has used quasiexperimental methods to study the differential effects of the California mandate across social groups. Kaiser provided members with free glucose test strips for 2 years after the legislation took effect, after which a 20% coinsurance was charged for test strip refills. The provision of free strips did not increase utilization, while the new coinsurance slightly decreased utilization, although not enough to have clinical relevance.*"[unreadable]'These effect sizes did not differ substantively across social groups. Harvard researchers previously studied the effect of a 1993 Harvard Pilgrim Health Care decision to provide free glucose meters as a member benefit. This policy resulted in increased utilization of SMBG and improved glycemic control overall,(117) but a less persistent adherence to SMBG practice in African Americans.(118) In contrast, there is also evidence that some health policies that improve population health could reduce rather than exacerbate health disparities. In a recent NEJM paper, investigators at the UCSF Center for Vulnerable Populations modeled the benefits of modest sodium reduction policies on cardiovascular disease outcomes (strokes and CVD-related death). These models suggest that health gains would be greater among African Americans and those of lower socioeconomic status. (119) The DREAM trial provides a recent example of clinical or biological differences by race/ethnicity in treatment effect. This 2x2 factorial, double-blind RCT was designed to evaluate the efficacy of thiazolidinediones on diabetes prevention among patients with non-diabetic dysglycemia. Not surprisingly, the authors reported substantial differences in the risk of diabetes onset by ethnicity, but also found a differential protective effect of thiazolidinediones by ethnicity.(120) Even if most trials observe no variation in outcomes across high risk subgroups (e.g., an observational effectiveness study at Kaiser showed no differential effect of initiating new diabetes medications (121), the uncertainty about racial or ethnic differences in effectiveness suggests a need for consistent testing in new populations.